Transient Activation of GABAB Receptors Suppresses SK Channel Currents in Substantia Nigra Pars Compacta Dopaminergic Neurons

Dopaminergic (DA) neurons in the substantia nigra pars compacta (SNc) are richly innervated by GABAergic neurons. The postsynaptic effects of GABA on SNc DA neurons are mediated by a mixture of GABAA and GABAB receptors. Although activation of GABAA receptors inhibits spike generation, the consequences of GABAB receptor activation are less well characterized. To help fill this gap, perforated patch recordings were made from young adult mouse SNc DA neurons. Sustained stimulation of GABAB receptors hyperpolarized SNc DA neurons, as previously described. However, transient stimulation of GABAB receptors by optical uncaging of GABA did not; rather, it reduced the opening of small-conductance, calcium-activated K+ (SK) channels and increased the irregularity of spiking. This modulation was attributable to inhibition of adenylyl cyclase and protein kinase A. Thus, because suppression of SK channel activity increases the probability of burst spiking, transient co-activation of GABAA and GABAB receptors could promote a pause-burst pattern of spiking.

While a number of studies have explored the excitatory glutamatergic regulation of SNc dopaminergic spiking [14,[19][20][21], far less attention has been paid to GABAergic synapses [18,22,23] despite roughly 70% of the synapses on SNc dopaminergic neurons being GABAergic [24,25]. Like most neurons in the brain, the effects of GABA on SNc dopaminergic neurons are mediated by ionotropic GABA A receptors and G-protein coupled GABA B receptors. In a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 spite of the fact that the GABA A receptor reversal potential is relatively depolarized [26] in SNc DA neurons, their activation clearly slows pacemaking [22,27]. However, the effects of synaptically released GABA on postsynaptic GABA B receptors has been more difficult to ascertain [14,28,29]. What is known is that stimulation of GABA B receptors with exogenous application of agonists leads to activation of K ir 3 K + channels, hyperpolarizing SNc dopaminergic neurons [30][31][32].
Although they provide a framework for understanding the effects of GABA on SNc dopaminergic neurons in vivo, these experiments have two limitations. First, the time course of receptor activation is much more prolonged than that expected to occur during a phasic burst of activity at GABAergic synapses. Second, much of the work has been conducted with a recording configuration that disrupts the oscillations in cytosolic Ca 2+ concentration that accompany pacemaking [33,34]. As intracellular Ca 2+ is a potent modulator of both plasma membrane ion channels and signaling pathways coupled to G-protein coupled receptors, this recording configuration might distort normal cellular responses.
To overcome these limitations, two experimental steps were taken. First, SNc dopaminergic neurons were recorded using the perforated patch technique with gramicidin D. This method only allows monovalent cations to pass between the cell and the electrode [35]. Second, GABA was applied transiently by optically releasing it from a chemical cage [36]. These experiments revealed that transient GABA B receptor stimulation preferentially suppressed SK K + channel currents by inhibiting constitutively active adenylyl cyclase and down-regulating protein kinase A activity. This suppression of SK K + channel currents induced a seconds long period of irregularity in the spiking of SNc dopaminergic neurons. Moreover, given previous work linking SK channels and burst spiking in SNc dopaminergic neurons, GABA B receptor signaling could promote burst spiking after a GABA A receptor induced pause.

GABA photolysis
RuBi-GABA [5μM] for uncaging experiments was purchased from Abcam (Cambridge, UK) and stock solutions were made and diluted within one week of use. Wide-field, 50 ms, 500 ms, and 1 min pulses from a CooLED (Andover, UK) pe-100 LED centered at 470nm (100% power) were used to uncage RuBi-GABA around the soma of the recorded cell.

2-Photon Laser Scanning Microscopy (2PLSM)
Brain slices were placed on an upright microscope heated chamber (~33˚C) and perfused at a constant rate of~2 mL/min with aCSF and allowed to stabilize for at least 15 mins before starting the acquisitions. Optical imaging of G-CEPIA1er signals were acquired using a 920-nm excitation beam (80-MHz pulse repetition frequency and~250-fs pulse duration), in a fixed plane of focus with a pixel size between 0.18 and 0.21 μm and a 12-μs pixel dwell time. The G-CEPIA1er fluorescence was detected by a GaSP PMT and a Dodt contrast-detector system that provided a bright-field transmission image (Prairie Technologies). Images were acquired with an Olympus LUMPFL 60×/1.0 NA water-dipping objective lens. One or two cell bodies were defined as a region of interest (ROI) for each experiment. Twenty frames of the G-CEPIA1er signal were collected in one optical plane at a rate of 3-4 frames per second. Acquisitions were taken every 10 mins for the baseline and the recovery phases, more frequently (1-5 mins) during stimulations.

Electrophysiology and imaging analysis
Data were analyzed using a custom written python analysis package (Neurphys; https://github. com/surmeierlab/neurphys) and MatLab scripts. Figures were created with Matplotlib [37] and Adobe Illustrator. All box plots presented as median, first and third quartiles, and whiskers at 10th and 90th percentiles. Data outside that range is represented as individual points.

TRAP tissue preparation
All experiments were approved by the Rockefeller University Institutional Animal Care and Use Committee and performed in accordance with the guidelines described in the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. Mice were housed in rooms on a 12 h dark/light cycle at 22˚C and maintained with rodent diet (Picolab) and water available ad libitum. 30 four-month-old male hemizygous Dat bacTRAP mice [38] were randomly divided into six groups of five mice. Brains were removed and sectioned using an icecold Adult Mouse Brain Slicer with 1 mm coronal slice intervals (Zivic Instruments). From the tissue section containing the midbrain, the SNpc and VTA regions were dissected and separated under a Nikon SMZ645 light microscope using a 10x lens.

RNA sequencing and analysis
Translated mRNAs were purified as described previously (Heiman et al. Nature Protocols 2014). TRAP samples underwent DNase digestion using the RNase-Free DNase Set (Qiagen) and were subsequently purified with the RNeasy MinElute Cleanup Kit (Qiagen). Eluted RNA samples were analyzed on a 2100 Bioanalyzer (Agilent) using RNA Pico Chips (Agilent) to confirm RNA integrity, followed by the measurement of RNA concentrations with the Quant-iT RiboGreen RNA Assay Kit (Life Technologies). cDNAs were prepared with the Ovation RNA-Seq System V2 kit (NuGEN), using an input of 1 ng RNA. 500 ng cDNA from each sample were fragmented on a Covaris S2 Focused Ultrasonicator using the operating conditions recommended by the manufacturer for a target fragment size of 200 bp. Fragment size was confirmed on a 2100 Bioanalyzer using High Sensitivity DNA Chips (Agilent). Libraries for RNA sequencing were prepared with the TruSeq RNA Sample Preparation v2 kit (Illumina), starting the manufacturer's low-throughput protocol with the end repair step. The concentration of the RNA-Seq libraries was determined on a 2100 Bioanalyzer using High Sensitivity DNA Chips. Subsequently, two libraries with different adapters were multiplexed for sequencing. After confirming the concentration of the multiplexed samples on a 2100 Bioanalyzer using High Sensitivity DNA Chips, samples were analyzed on an Illumina HiSeq 2000 sequencer using 100 bp single-end sequencing. RNA-Seq reads were mapped to the Mus musculus assembly 10 reference genome using TopHat version 2.0.11. FPKM values for all genes in each sample were calculated with Cufflinks version 2.2.1. To analyze differential gene expression between samples, DESeq version 1.14.0 was used under the standard comparison mode. P values were reported by DESeq, adjusted for multiple testing using the Benjamini-Hochberg procedure.

Transient stimulation of GABA B receptors has distinct effects
In ex vivo, coronal midbrain slices taken from P30-P45 mice, SNc DA neurons (Fig 1A) recorded using the perforated patch configuration exhibited regular, pacemaking activity (1-6 spikes/s) ( Fig 1B) [31,39-41]. When the Na V 1 channel antagonist tetrodotoxin (TTX, 1 μM) was added to the bath, slow oscillatory potentials (SOPs) appeared, which had a dominant frequency slower than that of pacemaking ( Fig 1B) [33,34, 42,43]. These SOPs were dependent upon opening of Ca V 1 Ca 2+ channels, as bath application of dihydropyridines at low micromolar concentrations eliminated them [33,34]. Bath application of the GABA A receptor antagonist gabazine (25 μM) did not alter either basal spiking rate or SOPs, suggesting that there was no tonic GABAergic tone with superfused slices (data not shown). As shown previously, bath application of the GABA B receptor agonist baclofen (5 μM) induced membrane hyperpolarization (Fig 1B). Sustained uncaging of Rubi-GABA (5 μM) in the presence of the GABA A receptor antagonist SR-95531 (gabazine, 25 μM) [44,45] also induced a consistent membrane hyperpolarization ( Fig  1D). These effects are in agreement with the previously described coupling of GABA B receptors through G i/o proteins to K ir 3 K + channels [30,46].
To determine if the duration of GABA B receptor activation had an impact on the physiology of SNc neurons, RuBi-GABA (5 μM) was uncaged for varying periods of time using fullfield LED illumination in the presence of gabazine (25 μM) to block GABA A receptors, and the effect on pacemaking monitored. A one minute uncaging pulse led to the expected slow hyperpolarization of the membrane and the eventual cessation of firing seen with bath application of  (Fig 2A). The onset of the hyperpolarization was slow, consistent with either a concentration dependence or a kinetically slow coupling mechanism. In contrast, a single 50 ms uncaging pulse caused a small, but consistent increase in discharge rate (Fig 2A) with a more pronounced increase in the irregularity of spiking as measured by interval standard deviation (Fig 2A). Within seconds, spiking returned to its normal rate and regularity. Bath application of the GABA B receptor antagonist CGP 55845 (2 μM) prior to GABA uncaging blocked changes in firing frequency and regularity (Fig 2A). Moreover, omission of the GABA A receptor antagonist gabazine during GABA uncaging did restore a brief delay in the next spike latency attributable to GABA A receptors; however, it did not qualitatively change the subsequent increase in discharge rate and irregularity (Fig 2C and 2D).
To directly assess the ability of GABA B receptor signaling to modulate SK channels, a hybrid clamp technique was used. In perforated patch recordings, a brief current pulse was used to evoke a spike and the cell was voltage clamped at -60 mV, allowed SK K + currents to be reliably measured [48]. The SK channel antagonist apamin almost completely eliminated the post-spike current (Fig 3A). Baclofen (5 μM) also dramatically reduced the post-spike current, providing direct evidence of SK channel modulation by GABA B receptors ( Fig 3A).
As mentioned above, SK channels have also been implicated in the regulation of the SOP [33,53]. In agreement with previous studies using sharp electrodes [53], in perforated patch recordings apamin (200 nM) caused the SOP to slow and become more irregular (Fig 3B).
GABA B receptor signaling inhibited adenylyl cyclase GABA B receptors couple to G i/o signaling pathways. In addition to activating K ir 3 K + channels, G i/o -protein coupled receptors can inhibit voltage-dependent Ca 2+ channels [54,55]. This type of modulation could mediate the GABA B receptor suppression of Ca 2+ activated SK channels. In agreement with previous work [48], an inhibitor of Ca V 3 (mibefradil, 10 μM) but not Ca V 1 (isradipine, 10 μM) Ca 2+ channels decreased SK channel currents measured with the hybrid clamp ( Fig 4A). However, there was no measureable effect of baclofen (5 μM) on Ca V 3 channel currents measured with a voltage step from -100 mV to -50 mV (Fig 4B). Although it is possible that incomplete voltage clamp obscured a modest modulation of these currents by GABA B receptor activation, the more plausible interpretation is that the SK channel modulation was mediated by some other mechanism.
To narrow the range of potential targets, an attempt was made to determine the signaling elements downstream of GABA B receptors. One of the best described targets of G i/o signaling is adenylyl cyclase (AC) [56,57]. SNc dopaminergic neurons express AC1 isoforms that are stimulated by Ca 2+ (S1 Table), raising the possibility that this form of AC is constitutively active during pacemaking. To test this hypothesis, the ability of GABA B receptors to modulate the SOP was examined before and after perturbing AC signaling. First, as shown above, a single, brief uncaging of GABA induced a slowing of SOP frequency and an increased irregularity in the oscillation (Fig 5A). Bath application of a membrane permeable analog of cyclic adenosine monophosphate (cAMP) (8-bromo-cAMP, 1 μM) blunted this modulation (Fig 5B and  5C). However, bath application of 8-bromo-cAMP (1 μM) had no discernible effect on SOPs (Fig 5D), consistent with the proposition that AC was constitutively active and GABA B receptors were having their effect by transiently suppressing this activity.
The cAMP generated by AC activates protein kinase A (PKA), leading to phosphorylation of a wide range of substrates. If GABA B receptor signaling was inhibiting AC, it should result in diminished PKA activity. Thus, PKA inhibition should mimic GABA B receptor signaling. To test this idea, the PKA inhibitor H-89 (10 μM) was bath applied. As predicted, H-89 slowed the SOP and increased its irregularity and amplitude (Fig 5E and 5G-5I). To directly test the role of AC signaling in regulating SK channels, the hybrid clamp was used. As expected, 8-bromo-cAMP had no effect on SK channel currents ( Fig 5H). However, H-89 dramatically suppressed SK channel currents (Fig 5H), arguing that constitutive AC and PKA activity was critical to maintaining SK activity and that GABA B receptor inhibition of AC could mediate SK channel inhibition. Similar results were seen with another PKA antagonist, Rp-8-CPT-cAMPS (100 μM) (Fig 5I).

Depletion of intracellular Ca 2+ stores did not block SK modulation
SK channels in SNc dopaminergic neurons also are known to be regulated by Ca 2+ released from intracellular stores controlled by inositol 1,4,5-triphosphate (IP3) receptors [58,59]. PKA  phosphorylation of IP3 receptors enhances their opening [60][61][62]. Thus, one potential way in which PKA might promote SK activation is by enhancing IP3 receptor mobilization of Ca 2+ . Depleting intracellular Ca 2+ stores by inhibiting the smooth endoplasmic reticulum Ca 2+ ATPase (SERCA) with cyclopiazonic acid (CPA, 10 μM) reduced SK channel currents measured with the hybrid clamp, in agreement with previous work [63]. However, in the presence of CPA, the PKA inhibitor H-89 (10 μM) continued to suppress SK channel currents (Fig 6A), suggesting that PKA was not constitutively effecting SK channels by enhancing IP3 receptor function.
To provide an additional test of this inference, the effects of PKA inhibition on the ability of group 1 metabotropic glutamate receptors (mGluRs) to deplete ER Ca 2+ stores was assessed. If PKA was constitutively enhancing IP3 receptor function, inhibiting PKA should blunt the ability of mGluRs to activate IP3 receptors and deplete ER stores. To test this prediction, group 1 mGluRs were activated with (S)-3,5-dihydroxyphenylglycine (DHPG, 10 μM). DHPG produced a robust lowering of ER Ca 2+ concentration measured in ex vivo brain slices with the genetically encoded ER Ca 2+ probe CEPIA1er [64]. However, H-89 (10 μM) failed to significantly diminish the DHPG evoked drop in ER Ca 2+ (Fig 6B), suggesting that PKA does not constitutively enhance IP3R-mediated Ca 2+ release in SNc dopaminergic neurons. Application of the ryanodine receptor (RyR) antagonist dantrolene (10 μM) had no effect on SK currents, suggesting that RyRs do not contribute to SK channel activation. Taken together, these results argue that PKA is regulating SK channel gating through a mechanism that is independent of ER stores (Fig 7).

Discussion
Our studies have identified a novel mechanism of GABA B receptor modulation of SNc dopaminergic neuron activity. While confirming that GABA B receptor signaling can activate K ir 3 K + channels and suppress spiking when stimulation is sustained, our results show that transient activation of GABA B receptors has a very different effect on ongoing pacemaking. Rather than inhibiting spiking, transient activation of GABA B receptors increased spiking rate and irregularity. This effect on spiking was mediated by suppression of SK channel currents. Given the well-established ability of SK channel inhibition to increase the propensity of SNc dopaminergic neurons to spike in bursts [19,51], a natural inference from our studies are that in vivo GABA B receptors signaling could create a seconds long window in which subsequent glutamatergic input could induce burst spiking more easily.

Transient elevation in GABA induced suppression of SK currents
Using bath application of ligands, previous studies of GABA B receptor effects on SNc dopaminergic neurons have repeatedly found that they are capable of activating K ir 3 K + channels, leading to hyperpolarization and the cessation of ongoing, autonomous pacemaking [30,31]. The ability of G βγ proteins released by stimulation of G i/o coupled receptors to increase the open probability of K ir 3 K + channels through a membrane delimited signaling pathway has been extensively characterized in both native and heterologous expression systems [65][66][67]. This is a robust modulation that is resistant to the alterations in intracellular environment brought about by whole cell or sharp electrode intracellular recording techniques. Using the perforated patch technique, which largely preserves the intracellular milieu, including the oscillation in intracellular Ca 2+ concentration that accompanies pacemaking [34,68], this GABA B receptor modulation also was evident with bath application of GABA B receptor agonists. Thus, there is no apparent negative regulator of K ir 3 K + channel modulation in SNc neurons that are pacemaking and have normal fluctuations in intracellular Ca 2+ .
However, with brief stimulation of GABA B receptors enabled by optical uncaging of GABA, there was little evidence of K ir 3 K + channel activation in most cells (Fig 5C). Rather, uncaging GABA evoked a transient increase in discharge rate and a more prolonged period of irregularity in spiking. Both effects were attributable to a suppression in the mAHP generated by SK K + channel currents. This was shown not only by inspection of the voltage-trajectory of the somatic membrane potential but also with use of hybrid clamp techniques that allowed spike generated outward currents to be isolated. In these experiments, the SK K + channel blocker apamin mimicked the effects of GABA B receptor agonists. Apamin also is known to increase the irregularity of the SOP created by blocking Na V 1 Na + channels in SNc dopaminergic neurons. Again, this effect was mimicked by GABA B receptor agonists.
In contrast the modulation of K ir 3 channels, the effect of GABA B receptors on SK channels appeared to be mediated by G αi inhibition of adenylyl cyclase. There are several observations consistent with this conclusion. First, mRNA profiling of SNc DA neurons demonstrated they robustly express AC1, a isoform of adenylyl cyclase that is stimulated by Ca 2+ /calmodulin [69,70]. As pacemaking SNc DA neurons have high levels of intracellular Ca 2+ [34], AC1 should be constitutively activated. Previous work with whole cell recording where this intracellular Ca 2+ oscillation was disrupted could have missed the GABA B receptor modulation of SK channels because AC1 activity was reduced. Second, bath application of a membrane permeable cAMP analog (bypassing AC1) blunted the GABA B receptor modulation of SK channels. Moreover, as expected if there was constitutive activity of AC1, the cAMP analogue had no effect in the absence of GABA B receptor stimulation. Third, inhibiting PKA, a major target of Application of CPA to empty ER Ca 2+ stores significantly reduced SK current (orange, n = 6, Wilcoxon signed rank test, p = 0.0313) and further application of 10 μM H89 to inhibit PKA activity further significantly reduced SK current (brown, n = 6, Wilcoxon signed rank test, p = 0.0313), but does not reduce it any further than H89 alone (Mann-Whitney U test, p = 0.1320). Inhibiting ryanodine receptors with 10 μM dantrolene did not change SK current (black, n = 4, Wilcoxon signed rank test, p = 1.00). (B) Left, an example experiment showing the effect of DHPG application on ER Ca 2+ levels. Right, summary of normalized data showing a small, but not significant change in induced ER Ca 2+ release after application of H89 (n = 4) compared to control (n = 7) recordings (Mann-Whitney U test, p = 0.2303).
doi:10.1371/journal.pone.0169044.g006 cAMP signaling, mimicked the effects of GABA B receptor signaling and occluded the effects of GABA B receptor activation. All three of these observations point to a simple signaling model (Fig 7).
What is unresolved is how PKA signaling enhances SK K + channel gating. Neither source of Ca 2+ involved in SK K + channel gating-Ca V 3 Ca 2+ channels and IP3 receptor sensitive, ER Ca 2+ stores [48,63]-appeared be affected by PKA inhibition. However, it is possible that our assays of PKA modulation of these targets was not sensitive enough. It is also possible that the SK channel itself is a target of PKA. SK2 and SK3 channels are expressed by SNc DA neurons [47,50] and both channels have serine/threonine phosphorylation sites [71,72]. Although these sites have been reported to control membrane trafficking, it is not clear whether they also affect channel gating.
Another unresolved question is why transient uncaging of GABA was effective in triggering modulation of SK channels but not K ir 3 channels. As the duration of GABA uncaging will affect the concentration of GABA achieved in the extracellular space, one possibility is that there are high and low affinity GABA B receptors that differentially couple to SK and K ir 3 channels, but there is no evidence for this kind of GABA B receptor heterogeneity. Another possibility is that differential scaffolding of targets in the neighborhood of GABA B receptors is responsible. It is possible that in SNc dopaminergic neurons the density of K ir 3 channels in the neighborhood of GABA B receptors is low and that G βγ subunits released by receptor binding have to diffuse a Schematic diagram depicting hypothesized signaling pathways involved in the GABA B receptormediated inhibition of SK channels. GABA B receptor inhibition of AC by G i signaling is hypothesized to be responsible for reduced cAMP levels and PKA signaling. The reduction in PKA activity is hypothesized to reduce SK channel opening through mechanism that are independent of either plasma membrane Ca 2+ channels or release from intracellular stores.
doi:10.1371/journal.pone.0169044.g007 substantial distance to interact with them, slowing the modulation. AC1, on the other hand, could be held near GABA B receptors by scaffolding proteins [55], allowing G i subunits to quickly inhibit enzymatic activity.
Could GABA B receptors contribute to burst spiking?
Based upon previous work identifying K ir 3 channels as targets of GABA B receptor signaling, it is widely assumed that GABA B receptors inhibit SNc DA neuron spiking much like GABA A receptors. Our results suggest an alternative scenario. In vivo, SNc DA neurons often follow a pause in spiking with a period of increased spiking or bursting [16,27]. This pattern of activity, particularly bursting, is thought to have profound effects on target structures like the striatum by transiently elevating extracellular dopamine concentration. This transient elevation has been linked to reward prediction errors as well as the initiation of movement [4,73].
It is widely held that burst spiking in SNc DA neurons is driven by excitatory glutamatergic synaptic activity [14]. Much like apamin [19], the transient suppression of SK K + channel currents by GABA B receptors should make it easier for glutamatergic synapse to drive bursting [19,20,23,74]. Thus, glutamatergic input to SNc DA neurons that temporally lagged a GABAergic volley from striatal spiny projection neurons that engaged GABA B receptors, would be very effective in evoking a burst of spikes. However, if the GABAergic input was maintained for a longer period of time, it could lead to engagement of K ir 3 K + channels and suppression of responsiveness to glutamatergic inputs.
Supporting Information S1 Table.